Light therapy for winter depression (Seasonal Affective Disorder, SAD), offers the promise of major alleviation of symptoms of fatigue, hyperphagia, hypersomnia, dysphoria, and anxiety that - according to recent surveys - are experienced by millions of people living in the middle-to- northern latitudes of the U. S., as well as a larger number of subsyndromal sufferers. Although early studies ha e consistently demonstrated rapid improvement in outpatient populations, individual protocols have been compromised by factors such as inadequate sample size, ambiguous controls, inconsistent level of severity at entry points into treatment and evaluation of relapse upon withdrawal, and application of lenient measures of symptom reduction. And although previous morning-evening comparisons of light scheduling have led to an impression of morning-light superiority, our preliminary data suggest that both are equally effective when presented as initial treatment. Evening light, however, becomes ineffective following morning treatment, with total lack of response observed thus far. Such evening-light decrement is also seen in reanalysis of cross-center data. This sequential dependency provides a procedural framework for distinguishing among alternate mechanisms that may subserve the specific action of light therapy. Our proposed experiments build upon a rigorously controlled protocol in which SAD patients receive successive, randomized tests of bright artificial light in morning and evening hours, including parallel-group controls designed to elucidate the morning-light carryover effect. We raise the new hypotheses, that (1) clinical response to light is effective at any time of day except when it induces circadian phase delay (as measured by nocturnal onset of melatonin secretion), and (2) light treatment serves to stimulate retinal photoreceptor metabolism, potentiating visual input signals to the central nervous system. We propose to test an interactive model of phase delay and retinal response through melatonin measurements and dark-adaptometry at selected points within the clinical protocol. Further, we plan to challenge these physiological explanations of light effect by addition of an inert placebo control. At issue is whether initial evening light response is nonspecific, and whether lack of response is induced by prior exposure to the putative active agent, morning light. Patients will also receive detailed ophthalmological evaluations pre-and posttreatment, and in long- term follow-up, in an ongoing effort to document ocular safety of bright light treatment, and to probe for the presence of any retinal effects.